I'm working on a school project to simulate an oscillating 2d fish centerline in fluent. The laminar flow model appears to produce adequate results, but all of the turbulence models seem to damp out the reverse Karmann vortex street that should be in the fish's wake, particularly in the Reynold's Stress model.

I have tried capping the turbulent viscosity ratio to lower than default and also reduced the initial values for turbulent kinetic energy and dissipation by three orders of magnitude only to get either similar or no results. I have initiated a couple of simulation runs using the k-e model in steady state for over 100,000 iterations to try to get better results with Reynold's Stress.

I can not use k-e for the dynamic run because it basically models what I am trying to ascertain. Does anyone know of any good way to reduce vortex damping caused by the turbulence models?

I know that in the spalart-allmaras there is a correction to acoount for vortical flows. You have to select the strain/vorticity based production option. This option is ruled by the coefficient cprod. The higher the coefficient, the stronger the correction. (papers advise cprod = 4).

Can't you use LES for this flow? (In 2D it can be activated with the text command (rpsetvar 'LES-2D? #t).

Tom,

Thank you so much! You are a lifesaver. I had no idea you could do 2D LES simulations in FLUENT. Again, thank you very, VERY much.

Tom,

Thank you so much! You are a lifesaver. I had no idea you could do 2D LES simulations in FLUENT. Again, thank you very, VERY much.

anyway be carefull with your results cause turbulence is 3d by definition. that's why les in 2d is not a common approach.